- Email: [email protected]
The PTP Family Photo Album
Leading Edge
Previews The PTP Family Photo Album Michel L. Tremblay1,*
Goodman Cancer Centre, Biochemistry Department, Montreal, Quebec H3G 0B1, Canada *Correspondence: [email protected] DOI 10.1016/j.cell.2009.01.006
1
Protein tyrosine phosphatases (PTPs) are central players in many biological processes. In this issue, Barr et al. (2009) analyze 22 different PTP structures to define their common and unique features. This effort provides key insights into the regulation of PTP activity that could lead to the development of new therapeutics. Phosphorylated tyrosine residues serve as unique and exquisitely sensitive switches for the initiation and termination of intracellular signals. Although cells have evolved several protein domains that recognize phosphotyrosine in the context of a wide number of proteins, the protein tyrosine phosphatases (PTPs) remain the most divergent and ubiquitously expressed gene families with this capacity. The human genome contains 107 PTPs, which recognize and remove phosphate moieties attached to tyrosine/serine/threonine containing protein substrates. Based on their primary sequence, only 38 of these enzymes appear to be uniquely specific to phosphotyrosine and this group is so divergent that they have been further subclassified into 17 subgroups of enzymes. This diversity likely reflects the richness of the interactions of PTPs with the cell signaling machinery. In this issue, Barr et al. (2009) examine 22 PTP structures and present several new insights into substrate recognition and catalytic activity that will be of interest to both the PTP community and those seeking to identify PTP modulators for the treatment of various diseases. In the 20 years since their discovery, much of what we have learned about the mechanism of action of the typical PTP catalytic domain comes from crystal structures of a relatively small subset of these proteins. The PTP catalytic domains of these enzymes are generally easily purified and obtained in very active extracts. And although they display poor substrate specificity in vitro, they are extremely specific in vivo. This was recently demonstrated for PTP1B, which recognizes only one of multiple tyrosine
phosphorylation sites on cortactin, a protein that regulates actin branching (Stuible et al., 2008). Therefore, to reveal their common features, as well as the uniqueness of the different subgroups of PTPs, an assessment of a larger pool of PTP structures has been needed. Barr et al. (2009) have now analyzed the variation in surface properties of 22 human PTP family members. In addition to the many individual particularities that are revealed about each, they present the discovery of a new atypical conformation for the catalytic (WPD) loop in which the loop is open even when the catalytic pocket is bound by substrate. They also assay a broad range of phosphopeptide substrates and model some of those peptides onto PTP structures. Two sets of findings particularly stand out from their work. First, the systematic comparison of these different PTPs allowed the authors to show that a secondary substrate-binding pocket, initially reported for PTP1B (Barford et al., 1994), can also be found in most other PTPs. The secondary substrate-binding site is separated from the catalytic pocket by a small gateway region, which consists of a direct groove between the catalytic pocket and the secondary substrate binding site that is flanked by two bulky residues. Barr and colleagues examined these two features in all of their PTP structures and were able to categorize most of their candidate PTPs into five subsets of enzymes based on whether those two structural elements were “open” or “closed.” For example, they observed that the secondary substrate-binding site could be anchored on a basic residue (Arg24 in PTP1B) or a cysteine for the R8 subgroup (IA2, IA2β).
Alternatively, other PTPs, such as STEP and HePTP, possess an aromatic or a proline residue at that site that eliminates the possibility of the region acting as a substrate-binding pocket. Moreover, they note that the gateway region is either “open” with sufficient space for a peptide to come into the groove or “closed” by the presence of bulky residues that prevent direct access between the secondary substrate-binding pocket and the catalytic pocket. The gateway region and the secondary substrate-binding site are likely to be crucial for substrate specificity. Importantly, PTP1B is already targeted by inhibitors that simultaneously obstruct the catalytic pocket and the secondary substrate-binding pocket (Zhang and Zhang, 2007). The authors’ recognition that most members of the PTP family have both a catalytic pocket and a secondary substrate-binding site provides an interesting and favorable rationale for the further development of such inhibitors. A second general feature of PTPs that is clarified by this new work is how homodimerization of receptor PTPs inhibits their activity. Receptor-like PTPs (RPTPs) comprise an extracellular segment, a transmembrane region, and an intracellular section containing one or two catalytic domains. In contrast to receptor tyrosine kinases that are activated by pairing, dimerization of RPTPs inhibits their activity. A mechanism for their inhibition through dimerization was first proposed by Bilwes et al. (1996) who noticed that RPTPα D1 crystals are indeed dimers. In their structure, a segment of 24 amino acids derived from a region between the transmembrane domain and the N terminus of each D1 domain, named the
Cell 136, January 23, 2009 ©2009 Elsevier Inc. 213
availability of a growing pool “wedge,” directly interacts of PTP structures is of great with the D1 catalytic pocket interest, and the 22 new strucof the partner protein. This tures concurrently published results in reciprocal steric hinby Barr and colleagues will drance of the catalytic pockhave an extraordinary impact ets (Figure 1). This model was on the daily efforts of a great supported by other work, such number of researchers workas by Majeti et al. (1998), who ing on PTPs. In particular, showed that a single mutathose seeking to develop new tion in this wedge domain PTP inhibitors stand to gain increases the in vivo activity of from this expanded toolbox of a hybrid EGFR-CD45. Xie et al. structures, bringing the broad (2006) also reported that pepclinical application of theratides derived from the wedge peutics that target PTPs one region of LAR (leukocyte antistep closer. gen-related protein tyrosine phosphatase) and PTPµ mediReferences ate homophilic interactions Figure 1. Dimerization Models of a Receptor Protein Tyrosine and reduce homodimeriza- Phosphatase. tion, resulting in an increase in (Left) Depiction of a (R5-like) monomeric active receptor protein tyrosine Barford, D., Flint, A.J., and phosphatase (RPTP) containing an extracellular segment with a carbonic anTonks, N.K. (1994). Science 263, PTP catalytic activity. hydrase motif (CA) and a fibronectin-like element (FN). The extracellular por1397–1404. Based on their study, Barr tion is linked through a transmembrane segment to the intracellular domain et al. (2009) reject the exis- composed of a proximal catalytic motif (D1) and a distal catalytic motif (D2). Barr, A.J., Ugochukwu, E., Lee, W.H., King, O.N.F., Filippakopou(Middle) An inactive PTP dimer is shown based on the “wedge” model of intence of such an inhibitory hibition that results from D1-D1 interactions. (Right) An inactive PTP dimer is los, P., Alfano, I., Savitsky, P., and wedge domain. In contrast shown based on the “head-to-toe” model of inhibition in which the D1 domain Burgess-Brown, N.A., Müller, S., and Knapp, S. (2009). Cell, this to the “wedge” model, the of each monomer interacts with the D2 domain of the other monomer. issue. RPTPγ crystal of the complete intracellular region reveals an inter- observed by Xie et al. (2006) still hint at a Bilwes, A.M., den Hertog, J., Hunter, T., and action between the D1 domain of one possible contribution of this intracellular Noel, J.P. (1996). Nature 382, 555–559. of the partners in the dimer with the D2 region to the dimerization process. Groen, A., Overvoorde, J., Van der Wijk, domain of the other. They recognize the The work of Barr et al. allows the exami- T., and den Hertog, J. (2008). FEBS J. 275, region between the D1 domain and the nation of a plethora of posttranslational 2597–2604. transmembrane domain not as a wedge modifications that control PTP activity, Majeti, R., Bilwes, A.M., Noel, J.P., Hunter, T., but as a linker segment that provides including phosphorylation, ubiquitination, and Weiss, A. (1998). Science 279, 88–91. sufficient flexible space to allow turning proteolytic cleavage, and oxidation. Oxida- Noordman, Y.E., Augustus, E.D., Schepens, of the full PTP internal domain by 90°. tion of PTPs, for example, causes various J.T., Chirivi, R.G., Ríos, P., Pulido, R., and HenTurning permits a head-to-toe interaction conformational changes that have a major driks, W.J. (2008). Biochim. Biophys. Acta 1783, with the other partner in the dimer. D1-D2 impact on enzyme activity. Indeed, Groen 275–286. recognition had also been observed by et al. (2008) recently confirmed that oxida- Stuible, M., Dubé, N., and Tremblay, M.L. (2008). Wallace et al. (1998), who used the D1 tion of RPTPs causes extensive conforma- J. Biol. Chem. 283, 15740–15746. domain of RPTPσ as bait in a two-hybrid tional changes in the cytoplasmic domains Wallace, M.J., Fladd, C., Batt, J., and Rotin, D. screen and identified the D2 domain of that stabilize the inactive state. Further (1998). Mol. Cell Biol. 18, 2608–2616. the close homolog RPTPδ as an interact- progress in crystallography will eventually Xie, Y., Massa, S.M., Ensslen-Craig, S.E., Maing partner. This D1-D2 interaction was lead to examination of the role of the trans- jor, D.L., Yang, T., Tisi, M.A., Derevyanny, V.D., considerably weakened by the removal membrane domain of receptor PTPs, such Runge, W.O., Mehta, B.P., Moore, L.A., et al. of the wedge domain from the D1 bait as the transmembrane domain of PTPRR (2006). J. Biol. Chem. 281, 16482–16492. construct. Yet this finding and the biolog- that is reported to promote the dimeriza- Zhang, S., and Zhang, Z.Y. (2007). Drug Discov. ical effects of wedge-derived peptides tion process (Noordman et al., 2008). The Today 12, 373–381.
214 Cell 136, January 23, 2009 ©2009 Elsevier Inc.
Previews The PTP Family Photo Album Michel L. Tremblay1,*
Goodman Cancer Centre, Biochemistry Department, Montreal, Quebec H3G 0B1, Canada *Correspondence: [email protected] DOI 10.1016/j.cell.2009.01.006
1
Protein tyrosine phosphatases (PTPs) are central players in many biological processes. In this issue, Barr et al. (2009) analyze 22 different PTP structures to define their common and unique features. This effort provides key insights into the regulation of PTP activity that could lead to the development of new therapeutics. Phosphorylated tyrosine residues serve as unique and exquisitely sensitive switches for the initiation and termination of intracellular signals. Although cells have evolved several protein domains that recognize phosphotyrosine in the context of a wide number of proteins, the protein tyrosine phosphatases (PTPs) remain the most divergent and ubiquitously expressed gene families with this capacity. The human genome contains 107 PTPs, which recognize and remove phosphate moieties attached to tyrosine/serine/threonine containing protein substrates. Based on their primary sequence, only 38 of these enzymes appear to be uniquely specific to phosphotyrosine and this group is so divergent that they have been further subclassified into 17 subgroups of enzymes. This diversity likely reflects the richness of the interactions of PTPs with the cell signaling machinery. In this issue, Barr et al. (2009) examine 22 PTP structures and present several new insights into substrate recognition and catalytic activity that will be of interest to both the PTP community and those seeking to identify PTP modulators for the treatment of various diseases. In the 20 years since their discovery, much of what we have learned about the mechanism of action of the typical PTP catalytic domain comes from crystal structures of a relatively small subset of these proteins. The PTP catalytic domains of these enzymes are generally easily purified and obtained in very active extracts. And although they display poor substrate specificity in vitro, they are extremely specific in vivo. This was recently demonstrated for PTP1B, which recognizes only one of multiple tyrosine
phosphorylation sites on cortactin, a protein that regulates actin branching (Stuible et al., 2008). Therefore, to reveal their common features, as well as the uniqueness of the different subgroups of PTPs, an assessment of a larger pool of PTP structures has been needed. Barr et al. (2009) have now analyzed the variation in surface properties of 22 human PTP family members. In addition to the many individual particularities that are revealed about each, they present the discovery of a new atypical conformation for the catalytic (WPD) loop in which the loop is open even when the catalytic pocket is bound by substrate. They also assay a broad range of phosphopeptide substrates and model some of those peptides onto PTP structures. Two sets of findings particularly stand out from their work. First, the systematic comparison of these different PTPs allowed the authors to show that a secondary substrate-binding pocket, initially reported for PTP1B (Barford et al., 1994), can also be found in most other PTPs. The secondary substrate-binding site is separated from the catalytic pocket by a small gateway region, which consists of a direct groove between the catalytic pocket and the secondary substrate binding site that is flanked by two bulky residues. Barr and colleagues examined these two features in all of their PTP structures and were able to categorize most of their candidate PTPs into five subsets of enzymes based on whether those two structural elements were “open” or “closed.” For example, they observed that the secondary substrate-binding site could be anchored on a basic residue (Arg24 in PTP1B) or a cysteine for the R8 subgroup (IA2, IA2β).
Alternatively, other PTPs, such as STEP and HePTP, possess an aromatic or a proline residue at that site that eliminates the possibility of the region acting as a substrate-binding pocket. Moreover, they note that the gateway region is either “open” with sufficient space for a peptide to come into the groove or “closed” by the presence of bulky residues that prevent direct access between the secondary substrate-binding pocket and the catalytic pocket. The gateway region and the secondary substrate-binding site are likely to be crucial for substrate specificity. Importantly, PTP1B is already targeted by inhibitors that simultaneously obstruct the catalytic pocket and the secondary substrate-binding pocket (Zhang and Zhang, 2007). The authors’ recognition that most members of the PTP family have both a catalytic pocket and a secondary substrate-binding site provides an interesting and favorable rationale for the further development of such inhibitors. A second general feature of PTPs that is clarified by this new work is how homodimerization of receptor PTPs inhibits their activity. Receptor-like PTPs (RPTPs) comprise an extracellular segment, a transmembrane region, and an intracellular section containing one or two catalytic domains. In contrast to receptor tyrosine kinases that are activated by pairing, dimerization of RPTPs inhibits their activity. A mechanism for their inhibition through dimerization was first proposed by Bilwes et al. (1996) who noticed that RPTPα D1 crystals are indeed dimers. In their structure, a segment of 24 amino acids derived from a region between the transmembrane domain and the N terminus of each D1 domain, named the
Cell 136, January 23, 2009 ©2009 Elsevier Inc. 213
availability of a growing pool “wedge,” directly interacts of PTP structures is of great with the D1 catalytic pocket interest, and the 22 new strucof the partner protein. This tures concurrently published results in reciprocal steric hinby Barr and colleagues will drance of the catalytic pockhave an extraordinary impact ets (Figure 1). This model was on the daily efforts of a great supported by other work, such number of researchers workas by Majeti et al. (1998), who ing on PTPs. In particular, showed that a single mutathose seeking to develop new tion in this wedge domain PTP inhibitors stand to gain increases the in vivo activity of from this expanded toolbox of a hybrid EGFR-CD45. Xie et al. structures, bringing the broad (2006) also reported that pepclinical application of theratides derived from the wedge peutics that target PTPs one region of LAR (leukocyte antistep closer. gen-related protein tyrosine phosphatase) and PTPµ mediReferences ate homophilic interactions Figure 1. Dimerization Models of a Receptor Protein Tyrosine and reduce homodimeriza- Phosphatase. tion, resulting in an increase in (Left) Depiction of a (R5-like) monomeric active receptor protein tyrosine Barford, D., Flint, A.J., and phosphatase (RPTP) containing an extracellular segment with a carbonic anTonks, N.K. (1994). Science 263, PTP catalytic activity. hydrase motif (CA) and a fibronectin-like element (FN). The extracellular por1397–1404. Based on their study, Barr tion is linked through a transmembrane segment to the intracellular domain et al. (2009) reject the exis- composed of a proximal catalytic motif (D1) and a distal catalytic motif (D2). Barr, A.J., Ugochukwu, E., Lee, W.H., King, O.N.F., Filippakopou(Middle) An inactive PTP dimer is shown based on the “wedge” model of intence of such an inhibitory hibition that results from D1-D1 interactions. (Right) An inactive PTP dimer is los, P., Alfano, I., Savitsky, P., and wedge domain. In contrast shown based on the “head-to-toe” model of inhibition in which the D1 domain Burgess-Brown, N.A., Müller, S., and Knapp, S. (2009). Cell, this to the “wedge” model, the of each monomer interacts with the D2 domain of the other monomer. issue. RPTPγ crystal of the complete intracellular region reveals an inter- observed by Xie et al. (2006) still hint at a Bilwes, A.M., den Hertog, J., Hunter, T., and action between the D1 domain of one possible contribution of this intracellular Noel, J.P. (1996). Nature 382, 555–559. of the partners in the dimer with the D2 region to the dimerization process. Groen, A., Overvoorde, J., Van der Wijk, domain of the other. They recognize the The work of Barr et al. allows the exami- T., and den Hertog, J. (2008). FEBS J. 275, region between the D1 domain and the nation of a plethora of posttranslational 2597–2604. transmembrane domain not as a wedge modifications that control PTP activity, Majeti, R., Bilwes, A.M., Noel, J.P., Hunter, T., but as a linker segment that provides including phosphorylation, ubiquitination, and Weiss, A. (1998). Science 279, 88–91. sufficient flexible space to allow turning proteolytic cleavage, and oxidation. Oxida- Noordman, Y.E., Augustus, E.D., Schepens, of the full PTP internal domain by 90°. tion of PTPs, for example, causes various J.T., Chirivi, R.G., Ríos, P., Pulido, R., and HenTurning permits a head-to-toe interaction conformational changes that have a major driks, W.J. (2008). Biochim. Biophys. Acta 1783, with the other partner in the dimer. D1-D2 impact on enzyme activity. Indeed, Groen 275–286. recognition had also been observed by et al. (2008) recently confirmed that oxida- Stuible, M., Dubé, N., and Tremblay, M.L. (2008). Wallace et al. (1998), who used the D1 tion of RPTPs causes extensive conforma- J. Biol. Chem. 283, 15740–15746. domain of RPTPσ as bait in a two-hybrid tional changes in the cytoplasmic domains Wallace, M.J., Fladd, C., Batt, J., and Rotin, D. screen and identified the D2 domain of that stabilize the inactive state. Further (1998). Mol. Cell Biol. 18, 2608–2616. the close homolog RPTPδ as an interact- progress in crystallography will eventually Xie, Y., Massa, S.M., Ensslen-Craig, S.E., Maing partner. This D1-D2 interaction was lead to examination of the role of the trans- jor, D.L., Yang, T., Tisi, M.A., Derevyanny, V.D., considerably weakened by the removal membrane domain of receptor PTPs, such Runge, W.O., Mehta, B.P., Moore, L.A., et al. of the wedge domain from the D1 bait as the transmembrane domain of PTPRR (2006). J. Biol. Chem. 281, 16482–16492. construct. Yet this finding and the biolog- that is reported to promote the dimeriza- Zhang, S., and Zhang, Z.Y. (2007). Drug Discov. ical effects of wedge-derived peptides tion process (Noordman et al., 2008). The Today 12, 373–381.
214 Cell 136, January 23, 2009 ©2009 Elsevier Inc.